Kalman Filter Fusion for Robust System State Estimation

The patent divides the estimation task into multiple independent Kalman filters, each with different setting parameters (e.g., different process noise covariance matrices Qk or measurement noise covariance matrices Rk). Each filter processes the same sensor data independently, and their results are subsequently fused. This segmentation allows each filter to represent different error scenarios or model assumptions, improving overall robustness while keeping individual filter complexity manageable.

The patent combines multiple Kalman filter estimation results through a fusion mechanism that integrates the outputs from all filters into a single final estimation. The fusion process weighs and combines the estimates from filters with different settings, effectively merging their strengths to produce a more reliable overall system state estimation that is robust to various error conditions.

The patent segments the configuration complexity by providing pre-defined setting options for each Kalman filter rather than requiring custom configuration. Each filter corresponds to a specific set of parameters (e.g., different noise covariance matrices) that can be selected without deep understanding of the underlying mathematics, making the system easier to operate while maintaining adaptability.

The patent changes the approach to parameter configuration by offering discrete, pre-validated parameter sets rather than continuous adjustment options. Users can select from predefined settings (e.g., different Qk or Rk matrices) that are optimized for specific scenarios, eliminating the need to manually tune numerous parameters while still adapting to different system behaviors.